Merge pull request #2416 from seek4science/publication-abstract-trim

Strip leading/trailing whitespace from publication abstracts

Author: fbacall

app/models/publication.rb

@@ -49,7 +49,7 @@ class Publication < ApplicationRecord
   has_one :content_blob, ->(r) { where('content_blobs.asset_version =? AND deleted=?', r.version, false) }, as: :asset, foreign_key: :asset_id
 
   explicit_versioning(:version_column => "version", sync_ignore_columns: ['license','other_creators']) do
-  acts_as_versioned_resource
+    acts_as_versioned_resource
     has_one :content_blob, -> (r) { where('content_blobs.asset_version =? AND content_blobs.asset_type =?', r.version, r.parent.class.name) },
             :primary_key => :publication_id,:foreign_key => :asset_id
 
@@ -58,7 +58,7 @@ def doi_uri
     end
 
     alias_method :doi_identifier, :doi_uri
-end
+  end
 
   belongs_to :publication_type
 
@@ -78,6 +78,8 @@ def doi_uri
 
   accepts_nested_attributes_for :publication_authors
 
+  auto_strip_attributes :abstract
+
   # Types of registration
   REGISTRATION_BY_PUBMED = 1
   REGISTRATION_BY_DOI    = 2

lib/tasks/seek_upgrades.rake

@@ -19,6 +19,7 @@ namespace :seek do
     update_rdf
     update_morpheus_model
     db:seed:018_discipline_vocab
+    strip_publication_abstracts
   ]
 
   # these are the tasks that are executes for each upgrade as standard, and rarely change
@@ -99,6 +100,21 @@ namespace :seek do
     end
   end
 
+  task(strip_publication_abstracts: [:environment]) do
+    puts 'Stripping publication abstracts...'
+    updated_count = 0
+    Publication.select(:id, :abstract).find_each do |publication|
+      if publication.abstract.present?
+        stripped = publication.abstract.strip
+        if stripped.length != publication.abstract.length
+          publication.update_column(:abstract, stripped)
+          updated_count += 1
+        end
+      end
+    end
+    puts "... updated #{updated_count} publications"
+  end
+
   private
 
   ##

test/unit/publication_test.rb

@@ -539,4 +539,49 @@ class PublicationTest < ActiveSupport::TestCase
     assert_includes publication.related_models, assay_model
     assert_includes publication.related_models, model
   end
+
+  test 'strips leading and trailing whitespace from abstract on save' do
+    publication = FactoryBot.build(:publication, abstract:
+      "  \n  This is an abstract with leading and trailing whitespace  \n  ")
+
+    User.with_current_user(publication.contributor) do
+      publication.save!
+    end
+
+    assert_equal 'This is an abstract with leading and trailing whitespace', publication.abstract
+  end
+
+  test 'handles nil abstract gracefully' do
+    publication = FactoryBot.build(:publication, abstract: nil)
+
+    User.with_current_user(publication.contributor) do
+      publication.save!
+    end
+
+    assert_nil publication.abstract
+  end
+
+  test 'strips whitespace from abstract when extracted from pubmed' do
+    publication_hash = {
+      'title' => 'SEEK publication',
+      'abstract' => "    An investigation into blalblabla  \n  ",
+      'journal' => 'The testing journal',
+      'pubmed' => '12345',
+      'doi' => nil
+    }
+    bio_reference = Bio::Reference.new(publication_hash)
+    publication = Publication.new(
+      title: 'Test',
+      projects: [FactoryBot.create(:project)],
+      publication_type: FactoryBot.create(:journal)
+    )
+
+    publication.extract_pubmed_metadata(bio_reference)
+
+    User.with_current_user(FactoryBot.create(:person).user) do
+      publication.save!
+    end
+
+    assert_equal 'An investigation into blalblabla', publication.abstract
+  end
 end